Structural characterization
The catalyst preparation and synthesis steps are illustrated in Fig. 1. Spent coffee ground-derived carbon (SCC-KU) was synthesized by pyrolysis with KOH and urea activation at 800 °C. SCC-KU cells were impregnated with ruthenium chloride and annealed at 700 °C (Ru@SCC-KU). As shown in Fig. 2a, the FE-SEM image revealed the hierarchical porous and interconnected morphology of Ru@SCC-KU, which may result in a high surface area and facile uniform dispersion of ruthenium nanoparticles (Ru NPs). No significant differences were observed between SCC-KU and Ru@SCC-KU, as shown in Fig. S1.
(a) FE-SEM image of Ru@SCC-KU. (b) TEM image of Ru@SCC-KU (inset Ru NPs dispersion). (c) HRTEM image of Ru@SCC-KU. (d) HAADF-STEM image of Ru@SCC-KU with (e, f, g) corresponding elemental mappings Ru, C, and O.
TEM was used to evaluate the particle size and specific morphology of the Ru NPs on the electrocatalysts. As shown in Fig. 2b, the TEM image displays the uniform distribution of Ru NPs on the carbon support. HR-TEM images revealed the presence of spherical Ru NPs distributed on the porous carbon texture, which formed particles with an average size of 4.31 nm. The d-spacing of the Ru NPs was determined to be 0.23 nm (Fig. 2c). The HAADF-STEM image is shown in Fig. 2d, and elemental mapping images are shown in Fig. 2e, f, and g, corresponding to the uniform dispersion of Ru, C, and O on the Ru@SCC-KU. TEM images of Ru@SCC-KU-6 and Ru@SCC-KU-8 revealed similar results to those for Ru@SCC-KU, but some agglomerations of NPs were observed for Ru@SCC-KU-6 (Fig. S2). Ru@SCC-KU-U and Ru@SCC-KU-N3 show poor uniform dispersion of Ru on the porous carbon support. Ru@CB has an onion-like amorphous carbon structure and more and fewer dots than the other materials1,30.
XRD and Raman spectroscopy were carried out to test the electrocatalyst properties. The XRD patterns of SCC-KU, Ru@SCG-KU, Ru@SCG-KU-8, and Ru@SCG-KU-6 are shown in Fig. 3a. Generally, broad peaks corresponding to planar graphite (002) and hexagonal structured carbon (100) structures are detected at 23° and 43°, respectively, in the spectra of the samples10,31. The sharpest diffraction peaks related to metallic ruthenium (Ru0) (JCPDS No 00-006-0663) were clearly observed, validating the existence of Ru NPs. The sharp peaks centered at 38.4, 42.1, 44.0, 58.3, 69.4, 78.4, 84.7, and 85.9° indicate the lattice planes of Ru0, corresponding to (100), (002), (101), (102), (110), (103), (112), and (201), respectively. The high-temperature annealed catalyst (Ru@SCC-KU-8) is sharper than the others, indicating more crystallinity, which validates the HR-TEM results. The peak intensity of Ru@SCC-KU annealed at 700 °C was lower than that of Ru@SCC-KU annealed at high temperature, indicating the formation of Ru clusters at higher temperatures32. The XRD patterns of the other catalysts are displayed in Fig. S3. The spectra of the catalysts revealed similar Ru peaks and general broad peaks for amorphous carbon structures. According to the TEM image results, Ru@CB and commercial carbon black exhibited more intense XRD peaks, confirming the existence of a defective structure rather than as-prepared coffee waste-derived carbons33. The commercial catalyst (Ru/C) may contain a more crystalline carbon structure, showing a sharp peak around 27°, and may come from oxidized ruthenium (RuO2), such as RuO2 (JCPDS No 01-088-0322)22.
(a) XRD patterns, (b) Raman spectra, and (c) XPS survey spectra of electrocatalysts. High-resolution XPS spectra of (d) Ru3p and (e) O1s of Ru@SCC-KU. (f) N2 adsorption and desorption isotherms (inset pore size distributions) of the prepared catalysts.
In Raman spectroscopy (Fig. 3b), two distinguishable peaks are centered at 1343 cm−1 (D band) and 1600 cm−1 (G band). The intensity ratio (ID/IG) of the D band (sp3) to the G band (sp2) is a standard method for estimating the graphitic degree and defects of the carbon lattice34. The intensity ratios (ID/IG) for SCC-KU, Ru@SCC-KU-8, Ru@SCC-KU-6, and Ru@SCC-KU were calculated to be 0.98, 0.97, 0.96, and 0.95, respectively, indicating that the Ru@SCC-KU catalysts have structured more crystalline than the parent carbon support (SCC-KU) due to high-temperature annealing and ruthenium impregnation. Additionally, broad peaks could be observed at approximately 1450 cm−1 and 1150 cm−1, corresponding to the presence of amorphous carbon and the coexistence of sp3 and sp2 hybridized carbon phases, respectively35. The Raman shifts of the other comparative catalysts displayed higher intensity ratios (ID/IG), as shown in Fig. S4, which may be attributed to the abundance of defects in the carbon framework. These results were in excellent agreement with the TEM and XRD analyses.
Moreover, X-ray photoelectron spectroscopy (XPS) was carried out to estimate the chemical components and oxidation states of the elements in the catalysts. As shown in Fig. 3c, the XPS survey spectrum revealed the presence of metallic Ru, oxygen, and carbon in the Ru@SCC-KU cells. The Ru3p high-resolution spectrum exhibited two peaks at 463.7 and 485.0 eV attributed to metallic ruthenium (Ru0), as indicated by Ru3p2/3 and Ru3p1/3 (Fig. 3d), confirming the absence of an apparent oxidation state in the Ru NPs22. In contrast, Ru@SCC-KU-U and Ru@SCC-KU-N3 exhibited two pairs of peaks corresponding to Ru0 and oxidized Ru (Fig. S5), which can be effects of additional nitrogen dopants, which can form Ru–N bonds in the carbon network. The high-resolution spectrum of C1s is shown in Fig. S6. The peaks at approximately 284.7, 286.2, 287.8, 289.7, and 291.8 eV were attributed to the C=C/C–C, C–O, C=O, C=O–C, and π − π* satellites, respectively, which is in good agreement with the coexistence of sp3 and sp2 hybridized carbon atoms and π − π* satellites and reconfirms the presence of C=C bonds36. The high-resolution O1s spectrum confirmed that Ru@SCC-KU exhibited significant peaks corresponding to C=O, C–O, and O=C–OH (Fig. 3e) and even to trace amounts of Ru–O, which may be attributed to Ru–O–C interactions due to the presence of oxygen functional groups in the carbon texture. This result is significant evidence of Ru–O–C bonding during the impregnation process because parent carbon (SCC-KU) is rich in oxygen functional groups. As comparative catalysts, Ru@SCC-KU-U and Ru@SCC-KU-N3 possess Ru–O, C–O, and C=O bonded structures (Fig. S7). Hence, the XPS analysis confirmed that the metallic Ru NPs on Ru@SCC-KU formed a strong bond (Ru–O–C) in the porous and amorphous carbon network, as indicated by the XRD and Raman results. According to the XPS analysis, Ru@SCC-KU contains negligible (0.49 at%) amounts of ruthenium and 94.5 and 5.01 at% C and O, respectively (Table S1). The Ru content was similar to that determined via ICP analysis (3.90 wt%) if converted to the value (at%).
Interestingly, the oxygen content decreased significantly (SCC-KU: 7.33 at% of O1s) due to the reaction with metal moieties during impregnation and annealing. Because of the optimal size of defects, oxygen functional groups on the carbon surface, and high surface area, SCC-KU may have a high adsorption capacity for metal cations in solution37. This may be the critical reason for the uniform dispersion and strong interaction of Ru–O–C in the as-synthesized Ru@SCC-KU catalyst. Such the formation of Ru–O–C networks can enhance the number of electrocatalytic active sites and their conductivity, preventing the aggregation of NPs, as reported in recent publications22,24,38. The components of Ru@SCC-KU-6 are similar to our focused catalyst Ru@SCC-KU. However, according to the XRD results, Ru@SCC-KU-8 has slightly greater Ru (0.51 at%) and lower oxygen (4.19 at%) contents than the other samples. Ru@SCC-KU-U and Ru@SCC-KU-N3 had lower carbon contents and higher oxygen, Ru, and nitrogen contents than the other nitrogen-added samples. According to the literature, the combination of nitrogen, carbon, and metallic Ru could be an efficient and high-durability catalyst for HER application by facilitating suitable interactions between Me and N–C4,39,40.
Nitrogen adsorption and desorption isotherms were used to study the specific surface area and pore structure of the electrocatalysts. As shown in Fig. 3f, SCC-KU has no hysteresis loops, similar to type I isotherms, which may have a micropore-dominated structure41. In contrast, Ru@SCC-KU exhibited a type IV isotherm with a hysteresis loop between 0.4 and 0.8 P/P0, indicating mesopore formation in the carbon texture42. Ru@SCC-KU has a greater mesopore volume than the other comparative samples (Fig. 3f, inset), which validates the enlargement of pore size and volume after Ru impregnation and annealing. Ru@SCC-KU and the parent carbon SCC-KU possess large BET surface areas of 1170.9 and 1243.9 m2 g−1, respectively, which are slightly reduced after annealing (Table S2). These results also confirm that the formation of Ru NPs did not block carbon pores and maintained a larger surface area. Thus, Ru@SCC-KU-N3 has a lower surface area and poor pore volume, which may be attributed to obstacles from the Ru–N complex and Ru NP clusters and the blockage of a porous morphology during the impregnation procedure. Hence, the increased specific surface area of the as-synthesized porous carbon (SCC-KU) could increase the number of Ru NP active sites and their conductivity by preventing the aggregation of metal species and providing a uniform distribution in the hierarchical porous structure43. Several efforts have been made to improve the HER activity of ruthenium catalysts on porous carbon supports19,22,23,24; however, few attempts have been made to use biowaste-derived carbon as a support20,25.
Electrochemical analysis
The electrocatalytic activity of the as-synthesized samples was examined through a three-electrode system in N2-saturated 1.0 M NaOH electrolyte at a constant 25 °C. The mass loading of the catalyst on the glassy carbon was ~ 0.612 mg cm−2. Figure 4a shows linear sweep voltammetry (LSV) plots at a scan rate of 5 mV s−1 and 1600 rpm with a rotating disk electrode (RDE) without iR compensation for the Ru@SCC-KU and control samples. We investigated the electrocatalytic activity of bench marketing catalysts, such as Pt/C (10 wt% platinum on graphitized carbon) and Ru/C (5 wt% ruthenium on carbon), as a reference. The Hg/HgO and NaOH (1.0 M) reference electrodes were calibrated against a reversible hydrogen electrode (RHE) (Fig. S8, Supplementary Note 1). The overpotential (η10) at a current density of 10 mA cm−2 is used as a standard evaluation parameter for electrocatalysts because it is equivalent to the 12.3% efficiency of a solar water-splitting device44. As expected, Pt/C and Ru/C exhibited high HER activities that needed 40.1 and 35.1 mV, respectively, to reach a current density of 10 mA cm−2. Ru@SCC-KU has excellent HER activity, outperforming commercial Pt/C and Ru/C, which have the lowest potential (η10 = 27.0 mV). Ru@SCC-KU-6 and Ru@SCC-KU-8 also outperformed Pt/C, showing η10 = 33.5 mV and η10 = 29.2 mV, respectively. This result is in good agreement with the XRD and Raman spectral results. Figure 4b summarizes the overpotentials and Tafel slopes of the as-synthesized and commercial catalysts, in which Ru@SCC-KU has better HER activity than the other materials. According to the Tafel equation, Tafel slopes were calculated from LSV polarizations45. As shown in Fig. 4c, Ru@SCC-KU had the lowest Tafel slope (58.4 mV dec−1) of any other catalysts. According to the lowest Tafel slope of Ru@SCC-KU, hydrogen evolution reaction may occured by Volmer-Heyrovsky reaction pathway (H2O + e− → OH− + Had; Had + H2O + e− → OH− + H2)18,46. In which, electrolysis reaction is fast due to high hydrogen bond strength (~ 65 kcal mol−1) of ruthenium.
(a) Linear sweep voltammetry (LSV) curves of HER in the alkaline medium (1 M NaOH), (b) the diagram of the overpotentials at the 10 mA cm−2 and Tafel slopes. (c) Tafel curves obtained from LSV of Pt/C 10%, Ru/C 5%, Ru@SCC-KU-6, Ru@SCC-KU-8, and Ru@SCC-KU. (d) Nyquist plots of samples at an overpotential of − 10 mV, (e) chronopotentionmetric curve of Ru@SCC-KU recorded at 10 mA cm−2 of current density for 24 h, and (f) the long-term stability of Ru@SCC-KU in alkaline electrolyte.
The LSV curves and Tafel slopes of the other comparative catalysts are shown in Fig. S9. Interestingly, comparative catalysts, such as Ru@SCC-KU-U (η10 = 35.5 mV, Tafel slope: 64.2 mV dec−1) and Ru@SCC-Z (η10 = 33.5, Tafel slope: 62.3 mV dec−1), exhibited good HER performance; these catalysts have higher activity than Pt/C but lower activity than Ru@SCC-KU. These results are related to the high specific surface area and ruthenium active sites of these catalysts. Ru@SCC-KU-N3 (η10 = 58.4 mV, Tafel slope: 64.2 mV dec−1) and Ru@CB (η10 = 99.4 mV, Tafel slope: 101.9 mV dec−1) showed poorer HER activity than the other materials. This result is similar to the research of Baek et al. on the outstanding HER activity of Ru NPs on graphene nanoplates47. However, in the case of Ru on nitrogen-doped support, the catalytic activity was significantly reduced because the active sites of the metal species may be blocked by Ru–N–C coordination27. The HER performance of carbon-based catalysts in alkaline electrolytes is summarized in Table S3 to compare our results22,23,24,25,26,48,49,50,51,52,53. This table shows that our results are better than or similar to those of previously reported catalysts. Here, Yuan’s research group reported the use of ruthenium oxide on a nitrogen-doped carbon matrix (RuO2/N–C), which exhibited good HER performance (η10 = 40 mV)52. As mentioned above, because Ru–N coordination may block some active sites on RuNPs, the RuO2/N–C catalyst exhibited lower activity than our synthesized catalyst (Ru@SCC-KU).
Electrochemical impedance spectroscopy (EIS) measurements revealed a lower charge transfer resistance and a smaller radius in the Nyquist curve of Ru@SCC-KU than in the other reference catalysts, indicating a faster charge during electrochemical cycling and superior electrode kinetics (Fig. 4d)22,54. The Nyquist plots were fitted to an equivalent circuit: R1(Q(R2(CR3))), as shown in Fig. S10a. R1 may correspond to solution resistance, R2 electrolyte, intermediate layer resistance, where charge transfer occurs, Q constant phase element, C capacitance, and R3 is the resistance of the barrier layer. Ru@SCC-KU has a substantially lower charge transfer resistance (R2) value (7.56 Ω) than the other samples, including Ru@SCC-KU-8 (10.35 Ω), Ru@SCC-KU-6 (8.83 Ω), Ru/C (11.43 Ω), and Pt/C (12.06 Ω). This lower resistance results from the favorable electron transfer between Ru and the carbon surface, which enables faster charge transfer and improved intrinsic catalytic activity. Additionally, the Nyquist curve radius of the catalysts increased in the following order: Ru@SCC-KU < Ru@SCC-KU-8 < Ru@SCC-KU-6 < Ru/C < Pt/C. As estimated before, the electrochemical activity significantly increased because of the uniform distribution of Ru NPs on the hierarchically porous carbon, which could increase the conductivity.
We determined the turnover frequency (TOF) of Ru@SCC-KU and commercial catalysts to determine the inherent electrocatalytic efficiency following previously reported methods (Supplementary Note 2)48,55. The TOF of Ru@SCC-KU was greater (Fig. S10b) than commercial catalysts (Pt/C and Ru/C). Precisely, the TOF of Ru@SCC-KU was estimated to be 0.44 H2 s−1 at 50 mV, which was higher than that of Ru/C (0.33 H2 s−1) and Pt/C (0.38 H2 s−1). These values confirm that Ru@SCC-KU is an excellent commercial catalyst and competitive HER activity with other reported electrocatalysts38,56.
Thiocyanate (SCN−) and ethylenediaminetetraacetic acid (EDTA) poisoning experiments were employed to determine the main catalytically active species in Ru@SCC-KU cells. EDTA is a complexing agent that selectively coordinates single atomic metal species. Moreover, thiocyanate can form complexes with single atoms and nanoparticles57. The poisoning concentration was 50 mM in 1.0 M NaOH electrolyte. As shown in the LSV curves (Fig. S11), there was almost no change after adding EDTA to the electrolyte compared with when no poisoning occurred. The HER catalytic activity decreases significantly in the case of thiocyanate poisoning, suggesting the presence of abundant nanoparticles in the catalyst. Thus, the results validate that Ru NPs are the dominant component as well as the main active sites for HER activity and agree with the previously discussed characterization analysis.
The electrochemical active surface area (ECSA) was tested by cyclic voltammetry (CV) at different scan rates to determine the specific HER activity of the catalyst (Fig. S12). The SCC-KU device had a lower capacitance, with a value of 55.1 mF cm−2. The Cdl value of Ru@SCC-KU was estimated to be 70.3 mF cm−2, indicating the highest ECSA, which is in good agreement with the finding that porous carbon with Ru NPs could have exposed active sites in the catalyst58. In contrast, the capacitance of Pt/C (34.8 mF cm−2) was lower than that of SCC-KU. Additionally, the ECSA and active site density of Ru@SCC-KU were calculated with ̴ 4.34 1013 sites per cm2 (Supplementary Note 3)59.
A cyclopotentiometric (CP) test was employed for 24 h at 10 mA cm−2 to evaluate the long-term stability of Ru@SCC-KU in an alkaline electrolyte. Figure 4e shows that, compared with Pt/C and other catalysts, Ru@SCC-KU exhibited excellent stability, with a slight decrease of ̴ 5.2% after a 24-h reaction (Fig. S13a). Figure 4f shows the LSV curves of Ru@SCC-KU before and after the CP test; there was almost no apparent loss, as it displayed only a 3.4 mV negative shift at 40 mA cm−2, which confirmed its high stability in an alkaline medium. In addition, after the long-term stability test, TEM images of the Ru@SCC-KU cells exhibited no significant differences in morphology (Fig. S14).
Moreover, acidic leaching is performed on the Ru@SCC-KU catalyst to check durability and recycling availability. After long-term stability testing, Ru@SCC-KU was leached into a 1.0 M HCl solution with constant stirring for 12 h and washed with distilled water until the sample was neutral. Then, Ru@CC-KU was dried at 105 °C for 12 h. Finally, the amount of leached Ru@SCC-KU was checked via an electrochemical test. Surprisingly, there were no apparent differences, such as a negligible decrease in the LSV curves (Fig. S13b) or no morphological changes in the TEM image (Fig. S14c). These results validate the superior stability, durability, and recyclability of Ru@SCC-KU compared to those of commercial catalysts.